Inputs
Inputs: daily US stock prices from third-party data providers and public SEC Form 4 filings from EDGAR. Each list uses only data available when it is sealed. Visitors see our lists and modelled results, never raw vendor data.
Quantitative trading intelligence, recorded before the outcome.
Every US trading day, Saetta AI screens about 887 eligible US-listed stocks with a fixed, rules-based method and seals its top 8 as the overnight list at about 15:40 New York, before the closing auction. It also runs insider-activity alerts from public SEC Form 4 filings. Every list and alert is time-stamped before its outcome is known, and the forward record below keeps every result, losses included.
Recorded before the outcome:every list is sealed and time-stamped before its entry auction and is never changed afterwards.
Losses kept:every session stays in the forward record, losing ones included.
Saetta AI does not hold your money or place trades. Nothing on this site is personal financial advice. Payments are off. Important information.
Overnight list
Replay of sealed Saetta AI output
| Rank | Stock | Move | Status |
|---|---|---|---|
| 1 | RTX | Sealed | |
| 2 | NI | Sealed | |
| 3 | ABVX | Sealed | |
| 4 | LNT | Sealed | |
| 5 | FE | Sealed | |
| 6 | ATI | Sealed | |
| 7 | PPL | Sealed | |
| 8 | DTE | Sealed |
892 eligible stocks, 8 ranked. Risk gate: half size.
First list in this replay.
SHA-256 b825fba7…8d17b4 chained to 43596e5d…a41d70
| Rank | Stock | Move | Status |
|---|---|---|---|
| 1 | RTX | Same | Sealed |
| 2 | NI | Same | Sealed |
| 3 | LNT | 1 place up | Sealed |
| 4 | ABVX | 1 place down | Sealed |
| 5 | FE | Same | Sealed |
| 6 | ATI | Same | Sealed |
| 7 | PPL | Same | Sealed |
| 8 | TRP | New | Sealed |
893 eligible stocks, 8 ranked. Risk gate: half size.
Left the list since 9 Oct: DTE.
SHA-256 024fba14…c27b7a chained to b825fba7…8d17b4
| Rank | Stock | Move | Status |
|---|---|---|---|
| 1 | RTX | Same | Sealed |
| 2 | NI | Same | Sealed |
| 3 | LNT | Same | Sealed |
| 4 | ABVX | Same | Sealed |
| 5 | FE | Same | Sealed |
| 6 | PPL | 1 place up | Sealed |
| 7 | ATI | 1 place down | Sealed |
| 8 | TRP | Same | Sealed |
888 eligible stocks, 8 ranked. Risk gate: half size.
No stock left the list since 12 Oct.
SHA-256 7757351f…a55f64 chained to 024fba14…c27b7a
- 8stocks on each sealed overnight list
- 887eligible US stocks screened for the latest list
- 15:40New York time each list is sealed, before the closing auction
- SHA-256each list chained to the one before
- 1 Oct 2026forward record start; never back-filled
- US$20kmodel account, after modelled IBKR costs
Live shadow sessionovernight, pre-market, regular session and after-hoursTue 6 Oct 2026
Engine test, Tue 6 Oct 2026 (retired test core)
A one-day shadow run of an earlier real-time core on the live US market feed, 03:59 to 20:00 New York. It served no customer, and that test core has since been shut down. The customer minute engine is a separate system and is not live yet.
fc2881e94b6f…
517,860,747
market events recorded, 03:59 to 20:00 New York, on 5 whole-market channels: trades, quotes, one-second bars, one-minute bars and limit-up/limit-down bands.
Live shadow session, 6 Oct 2026
- Symbols covered
- 18,489
with at least one event, on a whole-market subscription. 6,563 of them were outside the day's 11,932-symbol universe.
- The 1,000 most traded999 of 1,000
- The next 2,0001,999 of 2,000
- The other 8,9328,928 of 8,932
- Events per second
- 1,844 median · 32,963 p95
Peak: 272,099 in one second, at 10:04:29 New York. Mean 8,979.9.
Live shadow session, 6 Oct 2026 - Lag, exchange timestamp to engine core
- 136ms median · 204ms p95
Trades. Quotes: 136 ms median, 201 ms p95.
Live shadow session, 6 Oct 2026 - Segments sealed and verified
- 5,530 of 5,530
Each sealed journal segment carries its own SHA-256. Verified: it read back with exactly that hash.
Live shadow session, 6 Oct 2026 - Drops · reconnects · errors
- 0 · 2 · 0
Errors count malformed messages, feed error messages, bad segments and journal failures.
Live shadow session, 6 Oct 2026 - Core CPU
- 0.31 cores, mean
p95 0.67, peak 2.95, for the core's stream processes.
Live shadow session, 6 Oct 2026
Every figure on this panel: live shadow session, overnight, pre-market, regular session and after-hours, Tue 6 Oct 2026, 03:59 to 20:00 New York. Measured: live market-data ingest and the sealed journal; scoring was not attached. Not customer-facing. How it is verified.
How this is verified
Every figure above is computed from one sealed evidence folder for this session. Its manifest lists the SHA-256 of each of its 60 files. The site's generator checks every file against the manifest and refuses on any mismatch, missing file or extra file. It then checks the files against each other (the per-minute counts and the channels each add up to the total) and against the session report, and writes the panel's data file. The page adds no number of its own.
Evidence manifest SHA-256: fc2881e94b6f158e2610033090cee3c17b4ae9b9b72363438b1298b4fbb9abb7
The stream stopped cleanly and sealed every journal file. Up to 2 records still in flight after the stop signal were not written; they are not in any figure.
Lag runs from the exchange timestamp on each trade or quote to its receipt in the engine core, including the network path to the core's data centre, and carries the clocks' own small uncertainty. Shadow sessions run on the live market feed and serve no customer.
How Saetta AI works
From daily market data and public filings to a sealed list and a public forward record. The method is proprietary. Full page.
- Data:daily prices for US-listed stocks from third-party providers, and public SEC Form 4 filings.Daily inputs
- Rank:a fixed, rules-based ranking of the eligible stocks; the top 8 make the list.Fixed rules
- Seal:at about 15:40 New York the list is time-stamped and chained by SHA-256 to the one before.About 15:40 NY
- Hold:modelled entry in the closing auction, exit in the next opening auction.16:00 to 09:30 NY
- Record:the result joins the forward record once prices are final, losses included.After final prices
Data-driven stock rankings
Each overnight list ranks 8 stocks, 1 to 8, from the eligible US stocks that day. A rank is not a price target, and it does not say what you should do.
Ranks run from 1 to 8 on each overnight list. This strip is the scale, not a live ranking.
One overnight list, every US trading day
The daily schedule in New York time.
15:40: list sealed · 16:00: closing auction, modelled entry · overnight hold · 09:30: next opening auction, modelled exit · result added once prices are final
15:40: list sealed
About 15:40 New York
The daily overnight list of 8 stocks is sealed, time-stamped and chained by SHA-256 to the list before it.
16:00: closing auction, modelled entry
16:00 New York
Modelled entry in the closing auction for the overnight list, sized in whole shares after modelled costs.
Overnight hold
16:00 to 09:30 New York
The modelled overnight positions are held from the closing auction to the next US trading day's opening auction.
09:30: next opening auction, modelled exit
09:30 New York
Modelled exit in the next opening auction; result added once prices are final, losses included.
New York time, on US trading days. Exchange holidays and early closes follow the exchange calendar.
What's inside
Point-in-time inputs used for the sealed overnight lists and forward record. Visitors see our lists and modelled results, never raw vendor data.
Daily prices
Daily prices: opening and closing prices for US-listed stocks from third-party providers; used to rank and to price the forward record.
SEC Form 4 filings
SEC Form 4 filings: insider purchases reported to the SEC's EDGAR system; used for insider-activity alerts.
Sealed lists
Sealed lists: every overnight list, time-stamped and SHA-256-chained to the one before.
How a signal earns its place
Every signal set is tracked forward from its start date.
Each new signal set starts its own forward record on its start date. Records are never back-filled or joined. Machine-learning models have been tested in our research; none is used in anything we publish today.
One overnight pick, start to finish
Sealed: about 15:40 New York, time-stamped and chained · Entry: the closing auction (modelled) · Hold: overnight · Exit: the next opening auction, 09:30 New York (modelled) · Recorded: the result joins the forward record once prices are final.
- Sealed:about 15:40 New York, time-stamped and chained
- Entry:the closing auction (modelled)
- Hold:overnight
- Exit:the next opening auction, 09:30 New York (modelled)
- Recorded:the result joins the forward record once prices are final.
How it compares
Features of Saetta AI next to a typical retail signal service.
| Feature | Saetta AI | Typical retail signal service |
|---|---|---|
| Track record | Track record: a forward record from 1 October 2026, losing days kept | Often selected highlights |
| Research | Research: published, including where it made less than SPY | Rarely published in full |
| Integrity | Integrity: each overnight list SHA-256-chained to the one before (checked inside Saetta AI) | Rarely offered |
Security you can check
Sign-in, keys, chained lists and your money: what protects each one.
- Passkey or authenticator app: Sign in with a passkey, such as Face ID or Touch ID, or with an authenticator app. We do not use SMS codes.
- Keys stay in the header: Send a key only in the request's Authorization header, never in a web address. Keys expire, and you can revoke any of them at any time.
- Chained lists: Each overnight list carries the SHA-256 hash of the list before it. These checks run inside Saetta AI; visitors cannot verify them independently.
- Your money stays yours: We never hold money, connect to your broker or place a trade, and we never ask for a brokerage login.
- No trackers: No analytics or advertising cookies and no third-party trackers on this site. The portal sets one strictly necessary session cookie.
- Your account, your control: Sign-in history, sign out everywhere, recovery codes and a data download. Delete account disables the account and revokes its active keys.
Forward record
A record that starts on a date and only moves forward. Each session's signals are recorded and time-stamped before the outcome is known, and the modelled result is added once prices are final. Results are modelled, not live trades. Losing sessions, pending rows and sessions where a signal set did not run all stay in the record.
Forward record since 1 October 2026. Latest session: 14 October 2026. Updated 15 October 2026, 01:30 UTC.
The fair account: the same picks, with idle cash in SPY, against SPY
The next session is waiting for final data. The sessions below stay as recorded.
2Completed sessions, 12 October 2026 to 13 October 2026
−US$40.52Fair account net result after modelled costs
−US$23.54SPY over the same sessions
| Session | Result | Running total | SPY result | SPY running total |
|---|---|---|---|---|
| 13 Oct 2026 | −US$18.08 | −US$40.52 | −US$2.25 | −US$23.54 |
| 12 Oct 2026 | −US$22.44 | −US$22.44 | −US$21.29 | −US$21.29 |
A separate modelled account, not broker fills. It starts with US$20,000 at the 12 October 2026 opening auction, buys the same signals, and keeps cash the signals do not need in whole SPY shares, selling SPY when the signals need the money. Modelled IBKR Tiered costs and 2 basis points of slippage on each side apply. The SPY line starts at the same auction. Most of this account can be in SPY, so it can fall with the market. Every completed session stays in, losses included.
The model account: US$20,000, after modelled costs, next to SPY
9Market sessions tracked, 1 October 2026 to 13 October 2026
16Modelled orders
−US$87.62Model account net result after modelled costs
+US$350.53SPY reference over the same dates (not like for like; see below)
Session by session, latest first
| Session | Orders | Result | Running total | SPY running total |
|---|---|---|---|---|
| 13 Oct 2026 | 1 | −US$2.67 | −US$87.62 | +US$350.53 |
| 12 Oct 2026 | 1 | −US$15.55 | −US$84.95 | +US$352.87 |
| 9 Oct 2026 | 2 | −US$82.39 | −US$69.40 | +US$365.09 |
| 8 Oct 2026 | 1 | +US$14.96 | +US$12.99 | +US$244.45 |
| 7 Oct 2026 | 1 | −US$5.61 | −US$1.97 | +US$329.99 |
| 6 Oct 2026 | 2 | −US$46.42 | +US$3.64 | +US$378.61 |
| 5 Oct 2026 | 2 | +US$31.62 | +US$50.06 | +US$267.85 |
| 2 Oct 2026 | 4 | +US$5.38 | +US$18.44 | +US$132.91 |
| 1 Oct 2026 | 2 | +US$13.06 | +US$13.06 | −US$13.99 |
How it is modelled: each signal is recorded and time-stamped before its order time and is not changed afterwards. Orders are sized in whole shares for a US$20,000 model account and priced at the data vendor's daily opening and closing prices, after IBKR Tiered commissions with per-order minimums, modelled exchange, clearing and regulatory fees, and 2 basis points of slippage on each side. No order is sent to a broker. A session is added once its prices are final.
Comparison limits: the SPY reference holds the full US$20,000 in SPY from the opening price on 1 October 2026, while the model account starts at that day's closing price and puts only part of its money into each session's signals. Capital used and idle cash differ, so the two lines are not like for like.
Each signal set, session by session
9Completed dates
16Signals on completed dates
36.4%Signal-set results above zero
−28.2Average modelled result, basis points
| Session | Signal set | Status | Signals | Modelled result, basis points | Note |
|---|---|---|---|---|---|
| 14 Oct 2026 | Overnight list | Pending | 1 | — | Waiting for final prices. |
| 14 Oct 2026 | Insider-activity alerts | Pending | 1 | — | Waiting for final prices. |
| 13 Oct 2026 | Overnight list | Final | 1 | −22.1 | — |
| 13 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 12 Oct 2026 | Overnight list | Final | 1 | −127.8 | — |
| 12 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 9 Oct 2026 | Overnight list | Final | 1 | −104.5 | — |
| 9 Oct 2026 | Insider-activity alerts | Final | 1 | −139.8 | — |
| 8 Oct 2026 | Overnight list | Final | 1 | +120.0 | — |
| 8 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 7 Oct 2026 | Overnight list | Final | 1 | −46.0 | — |
| 7 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 6 Oct 2026 | Overnight list | Final | 2 | −193.2 | — |
| 6 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 5 Oct 2026 | Overnight list | Final | 2 | +134.5 | — |
| 5 Oct 2026 | Insider-activity alerts | Final | 0 | — | No signals were admitted. |
| 2 Oct 2026 | Overnight list | Final | 3 | −15.4 | — |
| 2 Oct 2026 | Insider-activity alerts | Final | 1 | +21.5 | — |
| 1 Oct 2026 | Overnight list | Final | 2 | +62.4 | — |
The overnight list is bought in the closing auction and sold in the next opening auction. Insider-activity alerts are bought in an opening auction and sold in a later closing auction, which can be in a later session. Alerts that were blocked before entry have no modelled result. The result is in basis points of the money the signals used; a basis point is one hundredth of one percent. Pending rows are not final yet, and rows marked not available stay in the record with their reason. The share above zero and the average use completed dates only; the share is of signal-set results, not of single trades.
These results are modelled from the engine's own signals, after modelled costs. They are hypothetical: not live trades and not a client account. Past performance is not a reliable indicator of future performance.
Saetta AI research results
Research accounts, 2017-2025, after all trading fees (IBKR Tiered, per-order minimums), next to holding SPY. The overnight ranking, with its settings re-chosen once a year using only data available at the time. Each button is one starting balance.
Slippage in these figures: 2 basis points per side.
SPY is an exchange-traded fund that tracks the S&P 500, an index of large US companies. SPY buy-and-hold is the same starting balance held in SPY over the same years, with dividends kept as cash.
Too small: below the US$2k margin minimum this research assumes, and it loses money after whole-share sizing and trading costs.
+US$290Net profit
US$5,290Ending balance
0.6%Growth per year (CAGR)
1.04Profit factor
−20.5%Max drawdown
- 2017−US$261SPY +US$1,024
- 2018−US$452SPY −US$260
- 2019US$0SPY +US$1,706
- 2020−US$53SPY +US$1,270
- 2021+US$281SPY +US$2,350
- 2022−US$23SPY −US$1,897
- 2023−US$16SPY +US$2,189
- 2024+US$1SPY +US$2,592
- 2025+US$813SPY +US$2,265
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January. The full-period net is the sum of these years.
SPY buy-and-hold (dividends kept as cash)
+US$11,239SPY net profit, same starting balance
US$16,239SPY ending balance
−32.0%SPY max drawdown
Average invested 16.7%. Trades 3,390.
+US$18,988Net profit
US$38,988Ending balance
7.7%Growth per year (CAGR)
1.17Profit factor
−21.1%Max drawdown
- 2017+US$1,900SPY +US$4,098
- 2018−US$946SPY −US$1,042
- 2019−US$1,767SPY +US$6,825
- 2020+US$8,073SPY +US$5,079
- 2021+US$2,546SPY +US$9,398
- 2022−US$1,318SPY −US$7,586
- 2023+US$671SPY +US$8,757
- 2024+US$4,142SPY +US$10,369
- 2025+US$5,687SPY +US$9,061
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January. The full-period net is the sum of these years.
SPY buy-and-hold (dividends kept as cash)
+US$44,959SPY net profit, same starting balance
US$64,959SPY ending balance
−32.0%SPY max drawdown
Average invested 42.0%. Trades 8,751.
+US$22,371Net profit
US$72,371Ending balance
4.2%Growth per year (CAGR)
1.10Profit factor
−29.0%Max drawdown
- 2017+US$6,020SPY +US$10,339
- 2018−US$190SPY −US$2,628
- 2019−US$4,939SPY +US$17,218
- 2020+US$7,795SPY +US$12,812
- 2021+US$6,235SPY +US$23,709
- 2022−US$12,883SPY −US$19,138
- 2023+US$1,490SPY +US$22,092
- 2024+US$7,953SPY +US$26,159
- 2025+US$10,889SPY +US$22,857
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January.
SPY buy-and-hold (dividends kept as cash)
+US$113,420SPY net profit, same starting balance
US$163,420SPY ending balance
−32.2%SPY max drawdown
Average invested 47.9%. Trades 14,488.
+US$92,267Net profit
US$192,267Ending balance
7.5%Growth per year (CAGR)
1.17Profit factor
−17.4%Max drawdown
- 2017+US$5,879SPY +US$20,678
- 2018+US$658SPY −US$5,257
- 2019−US$6,255SPY +US$34,436
- 2020+US$31,770SPY +US$25,624
- 2021+US$14,455SPY +US$47,418
- 2022−US$4,960SPY −US$38,277
- 2023+US$3,672SPY +US$44,184
- 2024+US$17,378SPY +US$52,319
- 2025+US$29,669SPY +US$45,715
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January.
SPY buy-and-hold (dividends kept as cash)
+US$226,840SPY net profit, same starting balance
US$326,840SPY ending balance
−32.2%SPY max drawdown
Average invested 46.6%. Trades 17,939.
+US$42,218Net profit
US$542,218Ending balance
0.9%Growth per year (CAGR)
1.03Profit factor
−42.7%Max drawdown
- 2017+US$37,951SPY +US$103,436
- 2018+US$11,457SPY −US$26,296
- 2019−US$14,764SPY +US$172,257
- 2020+US$130,905SPY +US$128,176
- 2021+US$27,662SPY +US$237,199
- 2022−US$60,212SPY −US$191,471
- 2023−US$113,674SPY +US$221,020
- 2024+US$32,967SPY +US$261,712
- 2025−US$10,074SPY +US$228,676
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January. The full-period net is the sum of these years.
SPY buy-and-hold (dividends kept as cash)
+US$1,134,709SPY net profit, same starting balance
US$1,634,709SPY ending balance
−32.2%SPY max drawdown
Average invested 64.9%. Trades 54,745.
+US$606,880Net profit
US$1,606,880Ending balance
5.4%Growth per year (CAGR)
1.11Profit factor
−30.4%Max drawdown
- 2017+US$81,448SPY +US$206,872
- 2018+US$20,521SPY −US$52,592
- 2019−US$22,662SPY +US$344,514
- 2020+US$273,113SPY +US$256,352
- 2021+US$66,346SPY +US$474,398
- 2022−US$66,992SPY −US$382,942
- 2023−US$166,094SPY +US$442,040
- 2024+US$221,577SPY +US$523,423
- 2025+US$199,624SPY +US$457,351
Green bars are up years. Red bars are losing years. Year figures are the same research account carried forward. They are not restarted each January.
SPY buy-and-hold (dividends kept as cash)
+US$2,269,418SPY net profit, same starting balance
US$3,269,418SPY ending balance
−32.2%SPY max drawdown
Average invested 69.7%. Trades 59,997.
Too large: a capacity limit blocks about half (52.3 percent) of the planned order value.
Backtested research results (hypothetical, after fees). Past performance is not a guarantee of future results.
The block above is backtested research. It is not a forward-tracked record.
How these figures were made: recomputed on 30 September 2026 with a stricter test (no look-ahead, costs and trade sizes modelled separately for each starting balance) and independently checked. Every balance shown made less than simply holding SPY over the same years; several had smaller drawdowns.
Still being built: not offered, no dates
These items are not offered to customers today, and no dates are given.
Minute rankings
Real-time minute rankings: being installed and tested; not offered yet.
Insider-activity alerts
Insider-activity alerts: run each trading day for the forward record; not offered to customers.
ASX
ASX: research only; no ASX signal is published.
Who it's for
Saetta AI publishes general research information for people who make their own decisions.
Not a fit if
You want someone to tell you what to buy, to manage money for you, or to promise an outcome. We do none of those. You can lose money.
What you get
Account management in the portal on this site.
Your account
Open an account in the portal on this site: email, password, country and an authenticator app; no SMS. New accounts are approved by hand during the evaluation, and data access is by invitation. Signed-in accounts can add a passkey, create, name and revoke API keys, and see their own usage. Log in.
Saetta AI pricing
Free during the evaluation. There are no paid plans, no prices and no checkout; do not send money. Prices and plan terms will be published on this page before any paid plan opens. Joining the waitlist does not reserve a price, a place or access. There is no uptime promise. The Terms say the service may change, pause or stop. Pricing page.
Evaluation access is by invitation and costs nothing. To ask about it, email hello@saettaai.com.
What this does not promise. No result, no level of accuracy, and no better outcome than the market. The 2017-2025 research backtest, after fees, made less than holding SPY at every supported account size on the home page. Several of those accounts had a smaller worst drop than SPY. The US$500k account had a larger one. No date is promised for anything marked coming soon. A forward record will be short at first. A few weeks of results cannot tell you whether a signal works. Past performance is not a reliable indicator of future performance.
FAQ
Plain answers. Open the FAQ page.
Is this financial advice?
No. Nothing on this site is personal financial advice. Information from Saetta AI is general and does not consider your objectives, financial situation or needs. Saetta AI does not hold an Australian financial services licence and is not registered as an investment adviser in the United States.
Where are the research results?
The home page shows a research backtest of the proprietary overnight ranking, 2017-2025, after fees, next to holding SPY, for each starting balance in the results file. A losing year stays on the chart. Open the results. The label on that block is backtested research results, and they are hypothetical. The home page also shows the forward record, which starts on 1 October 2026. Open the forward record. Past performance is not a reliable indicator of future performance.
Why show a backtest that made less than holding SPY?
Because that is what it showed. SPY is an exchange-traded fund that tracks the S&P 500, an index of large US companies, so holding it is a simple yardstick. Over 2017-2025, after fees, every supported account size on the home page made less than simply holding SPY. Several had a smaller worst drop than SPY. The US$500k account had a larger one. The research accounts were only partly invested, so they are not a like-for-like swap for an index fund. We would rather you see this now than find it later. Past performance is not a reliable indicator of future performance.
What is the difference between the backtest and the forward record?
A backtest applies rules to past data after the fact. The one on the home page covers 2017-2025, after trading fees. It is hypothetical, and it was built by people who already knew what had happened. A forward record starts on a date and only moves forward. From 1 October 2026, each overnight list and insider-activity alert is date-stamped when it is published, before the outcome is known. What happened next is recorded afterwards, by the same rules every time. Both are calculations from published signals. Neither is a client brokerage account. The forward record is on the home page.
When does the forward record start, and what goes in it?
It starts on 1 October 2026. It covers every overnight list and insider-activity alert from that date: losing sessions, pending rows and sessions where a signal set did not run stay in; alerts blocked before entry have no result. Entries are date-stamped when published, and we do not edit or remove them. If a model changes, the new version starts its own record from its own start date. Records are not joined together to make one look longer. A few weeks of results cannot tell you whether a signal works. Do not treat an early stretch, good or bad, as proof.
Will it tell me what to buy or sell?
No. Nothing here is a personal instruction. Saetta AI does not know what you own.
Can it trade for me or hold my money?
No. We are not a broker or a custodian. We never hold your money or investments, open trading accounts, or place, route or execute trades. Any trading decision, and its result, is yours.
Are payments on?
No. There are no prices and no checkout. Do not send money.
How do I create an account?
Open an account in the portal on this site: email, password, country and an authenticator app; no SMS. New accounts are approved by hand during the evaluation, and data access is by invitation. Signed-in accounts can add a passkey, create, name and revoke API keys, and see their own usage.
How is my account protected? Can I use a passkey?
Sign-up needs an email address, a password and an authenticator app. We do not use SMS codes. After you sign up, you can add a passkey, for example one that uses Face ID or Touch ID, and sign in with it. The Security page also lets you replace your recovery codes, see your sign-in history, and sign out everywhere. Passwords are not stored as the password itself. Account data is kept on a server in the Netherlands. We will never ask for your password or an API key by email. We never ask for brokerage logins, because Saetta AI does not connect to your broker.
Join the waitlist
Leave an email address to hear when access opens. Stock names and signals are not sent to the waitlist.